Wang Xingxing, Founder and CEO of Unitree Robotics
TMTPOST -- Wang Xingxing, founder and CEO of Unitree Robotics, said at the 2025 Inclusion Bund Conference on Thursday that while AI holds enormous potential to revolutionize robotics, the industry』s greatest challenges lie in securing high-quality data and developing robust models.
Speaking at a roundtable discussion, Wang argued that the field of applying AI to real-world tasks is still in its infancy, and success will depend on how effectively innovators bridge the gap between intelligent algorithms and the physical world.
He admitted that his biggest personal regret is not engaging with AI earlier, having focused exclusively on robotics since 2011. 「Currently, the field of getting AI to actually do work is still a barren desert—there may be a few blades of grass growing, but the real, large-scale explosive growth has yet to arrive,」 Wang said. The rapid advancement of AI now offers him a second chance to integrate intelligent systems into robotics, a move he believes could fundamentally reshape the industry.
This appearance marked Wang』s first public discussion since Unitree Robotics announced plans for an IPO. Founded in 2016, Unitree Robotics began with quadruped robots and is among the earliest companies globally to commercialize this technology. According to the GG Robotics Industry Research Institute, Unitree commanded 69.75% of the global quadruped robot market in 2023. The company also gained public attention during the Hangzhou Asian Games that year, when its quadruped robots delivered discus and javelins on the athletics field.
Unitree Robotics recently announced its intention to submit IPO application documents between October and December 2025. The company reported revenue exceeding 1 billion yuan in 2024, with profitability maintained for four consecutive years. Quadruped robots accounted for 65% of total revenue that year, highlighting the potential for consumer-level applications. If listed, Unitree would become the world』s largest publicly traded company specializing in both quadruped and humanoid robots, joining Zhiyuan Robotics as the second humanoid robotics firm on the public market. Public disclosure of R&D, customer data, and cost structures will also demonstrate that Unitree』s competitive edge extends beyond laboratory achievements.
Despite these successes, Wang stressed that the development of embodied intelligence and humanoid robotics is still in its early stages, with data remaining a key bottleneck. Large AI models benefit from massive digital datasets, yet data from the physical world is scarce. Pang Jiangmiao, head of the Embodied Intelligence Center at the Shanghai Artificial Intelligence Laboratory, echoed this point, noting that acquiring high-quality real-world data is critical for progress in robotics. Su Yang, co-founder and Chief AI Architect at Lingxin Qiaoshou, added that data from real devices is costly, but scaling to millions of machines could accelerate data collection and advance algorithms and computing infrastructure.
Earlier at the World Robot Conference on August 9, Wang had argued that the real challenge was not data, but suitable model architectures for embodied intelligence. He has since revised this view, acknowledging that both data and model quality are major challenges. 「I』m not saying data isn』t important. Previously, I meant that the industry focuses heavily on data issues but less on models. Right now, both data and models are core problems for robotics,」 he explained.
Wang also discussed organizational management, noting that Unitree』s rapid growth has created challenges in team collaboration and efficiency. In contrast, he highlighted that small teams in pure AI have the potential for explosive innovation if they possess top-tier talent. Despite the talent shortage, he remains optimistic, asserting that the AI era lowers barriers for young innovators and enables small teams to achieve outsized impact.
Looking forward, Wang envisions a robot capable of autonomous perception, decision-making, and interaction with the physical world—a true 「new species」 of AI. He stressed that understanding AI models as 「universal」 tools and embracing the present, rather than relying solely on past experience, are essential for breakthrough innovation.
「The AI era is incredibly fair—if you are smart and willing to act, even in a desert, towering trees will eventually grow,」 Wang said. He encouraged the younger generation to learn the latest knowledge and fully engage with the new era. 「We are still on the eve of AI truly being put to work and experiencing explosive, large-scale growth. For me personally, this is an incredibly exciting era.」
Wang』s insights reflect both the immense potential and the challenges facing the robotics industry as AI and physical automation converge. With Unitree Robotics preparing for its IPO, the company』s trajectory will likely set a benchmark for the next wave of AI-driven robotics innovation.
